Biology subjects

Lv, T.

Publications and source records attributed to Lv, T..

13 recordsLinked to original sources

bioRxiv2026

PatientPheX: A Full-Text Biomedical Corpus for Patient-Specific Phenotype Information Extraction

Patient-specific phenotype information extraction identifiers and normalizes phenotype entities and links them with individual patients, supporting rare disease research and genotype-phenotype analysis. However, existing datasets rarely combine full-text biomedical content with detailed annotations of patients, phenotypes, related entities, and their associations. In this work, we present PatientPheX, a corpus for patient-specific phenotype information extraction comprising 200 full-text articles, primarily case reports and case series. The corpus provides mention annotations and normalization for seven biomedical entity types, together with explicit associations between patients and entities across sentences, paragraphs, and sections. Negated findings and composite phenotype mentions are also annotated. Corpus construction combines automated pre-annotation, large language model (LLM)-assisted verification, and manual review and adjudication. Benchmarks cover entity recognition and normalization and patient-phenotype association extraction using gold-standard and predicted entities. The overall document-level entity F1 was 86.95%, while phenotype F1 was 79.33%. The best relation extraction macro-F1 decreased from 84.14% with gold-standard entities to 54.98% with predicted entities, highlighting the combined challenges of phenotype extraction and patient attribution. PatientPheX provides a resource for developing and evaluating patient-centered information extraction methods and supports the construction of structured patient-level knowledge resources for phenotype-driven biomedical research.

bioinformatics↗

bioRxiv2026

A unified spatial transcriptome profiling of ten mouse organs

Spatial transcriptomics has enabled numerous deep learning models in this area, and training them requires large amounts of high-quality data, especially expression matrices paired with histological images. Here, we present a unified spatial transcriptomic dataset generated using the Stereo-seq platform, covering 10 mouse organs--including brain, kidney, lung, thymus, large intestine, skin, spleen, ovary, testis, and uterus--encompassing 23 tissue sections generated from 21 chips, each with matched ssDNA or H&E staining images. The dataset comprises single-cell-resolution (cell-bin) or square bin-50 (25 {micro}m x 25 {micro}m) expression matrices for each sample, accompanied by corresponding cell type annotations. Annotation robustness was further supported by concordance across different sections of the same tissue and corroboration with canonical marker gene expression patterns. Finally, we compared the characteristics of the cell-bin and bin-50 expression matrices and demonstrated the advantages of cell-bin resolution for cell annotation. This dataset provides a standardized resource for spatial transcriptomics method development, benchmarking, and multimodal analysis.

bioinformatics↗

bioRxiv2026

Phytoplankton performance in the lab predicts occurrence in the field across a global temperature gradient

Biologists aim to predict where species will survive and thrive as the planet warms. To do so, we often rely on data-hungry species distribution models (SDMs) that use associations between species occurrences and environmental predictors to capture the realised niche. An alternative basis for predictions is to experimentally quantify the effect of environmental drivers on performance, which captures the fundamental niche. We presently do not know which of these approaches represents a better path towards accurate forecasts. SDMs may depend too strongly on present-day environmental covariation, which will change in the future. In contrast, a major shortcoming of experiments is that they ignore most environmental drivers to focus on one or two. Quantifying how well fundamental and realised niches agree today would help establish how useful both SDMs and experiments are likely to be. We therefore compared both niches in 39 relatively common marine phytoplankton species. The temperature-dependence of population growth rate was characterised with a thermal performance curve model applied to lab experimental data, and the temperature-dependence of species occurrence probability estimated with SDMs applied to a global compilation of marine presence records. We found a fairly strong, near 1:1 relationship between measures of thermal niche centre: the median growth temperature in the lab and the median occurrence temperature in the field (R2 = 0.49). We also found a modest positive relationship between measures of thermal niche width, the growth niche width and the occurrence niche width (R2 = 0.24). This agreement should increase our confidence in environmental preferences inferred with SDMs. It also suggests that simple experiments can reliably constrain species ranges and help forecast range shifts. This has important implications for forecasting community composition and ecosystem processes, as we ought to be able to predict range shifts in biogeochemically-important taxa such as diatoms and nitrogen-fixing cyanobacteria.

ecology↗

bioRxiv2026

PRADA-DTI: A Prototype-Retrieval Augmented Domain-Adaptation Framework for Drug-Target Interaction Prediction

Drug target interaction (DTl) prediction is a fundamental task in computational drug discovery. However, most existing DTI models assume a static learning environment, whereas real-world biomedical data are dynamic, characterized by the continuous emergence of new protein families and interaction patterns. This poses major challenges for model generalization and continual adaptation, especially under privacy and data-access constraints. To address these issues, we introduce PRADA-DTI, a retrieval-augmented, parameter-efficient framework for domain-incremental DTI learning. Built on a shared physicochemical backbone, PRADA-DTI learns domain-specialized prompts for representation-level modulation, coupled with Low-Rank Adaptation (LoRA) modules for parameter-space adaptation. At inference, protein embeddings query a compact prototype memory without storing raw molecular data, retrieving similar domains to dynamically compose the relevant prompts and LoRA parameters without requiring domain labels. This retrieval-guided composition enables continual learning from new protein domains while mitigating catastrophic forgetting on previous ones. On BindingDB and BIOSNAP benchmarks, PRADA-DTI substantially outperforms state-of-the-art continual learning and parameter-efficient baselines in both predictive accuracy and forgetting mitigation with minimal parameter updates. Interpretability analysis through residue-level attribution visualization demonstrates that the model correctly attends to binding pocket regions across different protein domains, confirming that the retrieval-guided adaptation mechanism captures biologically relevant structural patterns. These results demonstrate the effectiveness of retrievalaugmented parameter adaptation for continual drug discovery.

bioinformatics↗

bioRxiv2025

scProtoTransformer: Scalable Reference Mapping Across Molecules, Cells and Donors

The rapid accumulation of single-cell data has made it possible to comprehensively characterize biological systems at molecular, cellular, and donor levels. However, scalable reference mapping across different resolutions remains a major challenge in current research. Here, we propose scProtoTransformer, a prototype-based Transformer architecture designed to achieve scalable reference mapping across molecular, cell, and donor levels. scProtoTransformer introduces a knowledge-guided prototype tokenizer that projects gene expression into biologically interpretable pathway prototypes, effectively reducing numerical batch effects while preserving biological semantic patterns. Furthermore, by leveraging knowledge distilled from the foundation model and a dynamic supervised fine-tuning strategy, scProtoTransformer achieves robust biological representations with reduced pre-training requirements. Benchmark experiments across molecular, cell, and donor-level reference mapping demonstrate that scProtoTransformer delivers competitive or even superior performance compared with state-of-the-art approaches, while providing interpretability through biologically prototypes. Together, these results establish scProtoTransformer as a unified framework for scalable reference mapping, laying the foundation for systematic understanding from genes to individuals.

bioinformatics↗

bioRxiv2025

Cross-Species Prediction of Histone Modifications in Plants via Deep Learning

The regulation of gene expression in plants is governed by complex interactions between cis-regulatory elements and epigenetic modifications such as histone marks. While deep learning models have achieved success in predicting regulatory features from DNA sequence, their cross-species generalizability in plants remains largely unexplored. In this study, we systematically evaluated the ability of deep learning models to predict histone modifications across plant species using a multi-stage framework based on the Sei architecture. We trained species-specific models for Arabidopsis (A. thaliana), rice (O. sativa), and maize (Z. mays), achieving high within-species predictive performance (AUROC > 0.94) and strong agreement between predictions and experimental ChIP-seq profiles. However, cross-species predictions showed reduced performance with increasing phylogenetic distance, highlighting limited model transferability between monocots and dicots. To improve generalization, we built family-level models by training on multiple species across the Poaceae and Brassicaceae families. These models demonstrated robust predictive power in completely unprofiled species--those entirely unused in training set--highlighting the models adaptability to novel plant genomes based solely on conserved regulatory syntax. In contrast, cross-family models produced inconsistent results, with reliable performance only in species sharing conserved regulatory features. Lastly, we developed an easy-to-use pipeline that predicts genome-wide chromatin signals directly from DNA sequences. Our findings demonstrate that phylogenetically informed model training significantly improves cross-species epigenomic prediction, offering a scalable computational strategy for functional annotation in non-model and agriculturally important plants.

bioinformatics↗

bioRxiv2024

Single-cell multi-omics and spatial multi-omics data integration via dual-path graph attention auto-encoder

Single-cell multi-omics data integration enables joint analysis of the resolution at single-cell level to provide comprehensive and accurate understanding of complex biological systems, while spatial multi-omics data integration is benefit to the exploration of cell spatial heterogeneity to facilitate more diversified downstream analyses. Existing methods are mainly designed for single-cell multi-omics data with little consideration on spatial information, and still have the room for performance improvement. A reliable multi-omics data integration method that can be applied to both single-cell and spatially resolved data is necessary and significant. We propose a single-cell multi-omics and spatial multi-omics data integration method based on dual-path graph attention auto-encoder (SSGATE). It can construct neighborhood graphs based on single-cell expression data and spatial information respectively, and perform self-supervised learning for data integration through the graph attention auto-encoders from two paths. SSGATE is applied to data integration of transcriptomics and proteomics, including single-cell and spatially resolved data of various tissues from different sequencing technologies. SSGATE shows better performance and stronger robustness than competitive methods and facilitates downstream analysis.

bioinformatics↗

bioRxiv2023

Dendrimeric DNA Coordinate Barcoding Design for Spatial RNA Sequencing

Spatially resolved transcriptomic technologies show promise in revealing complex pathophysiological processes, but developing sensitive, high-resolution, and cost-effective methodology is challenging. Here, we report a dendrimeric DNA coordinate barcoding design for spatial RNA sequencing (Decoder-seq). This technology combined dendrimeric nano-substrates with microfluidic coordinate barcoding to generate high-density spatial DNA arrays with deterministically combinatorial barcodes in a resolution-flexible and cost-effective manner ([~]$0.5/mm2). Decoder-seq achieved high RNA capture efficiency, [~]68.9% that of in situ sequencing, and enhanced the detection of lowly expressed genes by [~]five-fold compared to 10x Visium. Decoder-seq visualized a spatial single-cell atlas of mouse hippocampus at near-cellular resolution (15 m) and revealed dendrite-enriched mRNAs. Application to renal cancers dissected the heterogeneous tumor microenvironment of two subtypes, and identified spatial gradient expressed genes with the potential in predicting tumor prognosis and progression. Decoder-seq is compatible with sensitivity, resolution, and cost, making spatial transcriptomic analysis accessible to wider biomedical applications and researchers.

genomics↗

bioRxiv2021

High-resolution spatiotemporal transcriptomic maps of developing Drosophila embryos and larvae

Drosophila has long been a successful model organism in multiple fields such as genetics and developmental biology. Drosophila genome is relatively smaller and less redundant, yet largely conserved with mammals, making it a productive model in studies of embryogenesis, cell signaling, disease mechanisms, etc. Spatial gene expression pattern is critical for understanding of complex signaling pathways and cell-cell interactions, whereas temporal gene expression changes need to be tracked during highly dynamic activities such as tissue development and disease progression. Systematic studies in Drosophila as a whole are still impeded by lack of these spatiotemporal transcriptomic information. Drosophila embryos and tissues are of relatively small size, limiting the application of current technologies to comprehensively resolve their spatiotemporal gene expression patterns. Here, utilizing SpaTial Enhanced REsolution Omics-sequencing (Stereo-seq), we dissected the spatiotemporal transcriptomic changes of developing Drosophila with high resolution and sensitivity. Our data recapitulated the spatial transcriptomes of embryonic and larval development in Drosophila. With these data, we identified known and previously undetected subregions in several tissues during development, and revealed known and potential gene regulatory networks of transcription factors within their topographic background. We further demonstrated that Stereo-seq data can be used for 3D reconstruction of Drosophila embryo spatial transcriptomes. Our data provides Drosophila research community with useful resources of spatiotemporally resolved transcriptomic information across developmental stages.

developmental biology↗

bioRxiv2021

Repression of MAPK/Erk signaling by Efnb2-Ephb4-Rasa1 is required for lymphatic valve formation

The lymphatic vascular system plays important roles in various physiological and pathological processes, and lack of lymphatic or lymphovenous valves always causes lymph or blood reflux, and can lead to lymphedema. However, the molecular mechanism underlying the valve formation is poorly understood. Here we report that the MAPK/Erk signaling needs to be repressed during the valve-forming lymphatic endothelial cells (LECs) fate determination, which differs from its positive role in the LECs specification. Up-regulation of MAPK/Erk signaling in ephb4b, efnb2a;efnb2b and rasa1a;rasa1b mutants leads to lymphatic valve defects, whereas simultaneous loss of Erk1 and Erk2 causes valve hyperplasia. Moreover, valve defects in ephb4b or rasa1a;rasa1b mutants are mitigated in the presence of MEK inhibitors, indicating a new function of Efnb2-Ephb4-Rasa1 cassette in lymphatic valve progenitor cells specification by repressing MAPK/Erk activity. Therefore, our findings provide a mechanistic understanding of the lymphatic valve formation and potential drug targets for related lymphatic diseases.

developmental biology↗

bioRxiv2020

Recapitulation of SARS-CoV-2 Infection and Cholangiocyte Damage with Human Liver Organoids

The newly emerged pandemic coronavirus, SARS-CoV-2, has posed a significant public health threat worldwide. However, the mode of virus transmission and tissue tropism is not well established yet. Recent findings of substantial liver damage in patients and ACE2+ cholangiocytes in healthy liver tissues prompted us to hypothesize that human liver ductal organoids could serve as a model to determine the susceptibility and mechanisms underlining the liver damage upon SARS-CoV-2 infection. By single-cell RNA sequencing, we found that long-term liver ductal organoid culture preserved the human specific ACE2+ population of cholangiocytes. Moreover, human liver ductal organoids were permissive to SARS-CoV-2 infection and support robust replication. Notably, virus infection impaired the barrier and bile acid transporting functions of cholangiocytes through dysregulation of genes involved in tight junction formation and bile acid transportation, which could explain the bile acid accumulation and consequent liver damage in patients. These results indicate that control of liver damage caused directly by viral infection should be valued in treating COVID-19 patients. Our findings also provide an application of human organoids in investigating the tropism and pathogenesis of SARS-CoV-2, which would facilitate novel drug discovery.

cell biology↗

bioRxiv2020

Meta-analysis of the effects of sleep deprivation on depression in patients and animals

ObjectiveResearch on the antidepressant effects of sleep deprivation (SD) is lagging and has not produced completely uniform results in humans and animals. The present study aimed to reassess the effect of SD on patients and animals by meta-analysis based on updated research. MethodsWe searched PubMed, Embase and Cochrane Library for articles since the first relevant literature published up to June 10th, 2019. Data on sample characteristics, features of SD, and tests for depression were extracted. ResultsFourteen articles were included, eight on humans and six on animals. We found that when the duration of SD in patients was 7-14 days, it reflected antidepression [-1.52 (-2.07, -0.97); I2=19.6%]. In animals, the results of sucrose consumption experiments showed that SD has depressogenic effects [-1.06 (-1.63, -0.49); I2=81.1%], while the results of forced swimming experiments showed that SD treated depression [-1.17 (-2.19, -0.16); I2=80.1%], regardless of the duration of sleep deprivation. ConclusionSD can be an effective antidepressant measure when the duration is 7-14 days in patients. In animal studies, SD has shown more antidepressant effects when measured by forced swimming experiments, whereas using sucrose consumption tests had the effect of worsening depression.

neuroscience↗

bioRxiv2019

Development of SSR markers related to seed storability traits in maize subjected to artificial seed aging conditions

Seed storability is an important and complex agronomic trait in maize because annual seed production considerably exceeds consumption. The viability of seeds decreases over time, even when stored at low temperature, until seeds finally lose viability. In our previous study, two inbred lines with significantly different storability, Dong156 with high storage tolerance and Dong237 with low storage tolerance, were selected over six years using a natural seed aging test. In the present study, an F2:3 population and a RIL (recombinant inbred line) population were constructed from these two inbred lines and used to map QTL (quantitative trait loci) with SSR (simple sequence repeat) markers. A phenotypic index of traits related to seed storability that includes germination rate, germination potential, a germination index, a vigor index, seedling weight, and seedling length was generated using the results of an artificial aging treatment. Two consistent regions, cQTL-7 on chromosome 7 and cQTL-10 on chromosome 10, were identified by comparing QTL analysis results from these two populations. After genotyping SSR markers in these two regions, cQTL-7 was remapped to between umc1671 and phi328175 in a 7.97-Mb region, and cQTL-10 was remapped to between umc1648 and phi050 in a 39.15-Mb region. Four SSR markers linked to cQTL-7 and cQTL-10, including umc1671, phi328175, umc1648, and phi050, were identified using a Chi-squared test. The combined selection efficiency of these four markers was 83.94% in 85 RIL lines with high storability, and marker umc1648 exhibited the highest efficiency value of 88.89%. These results indicated that the four SSR markers developed in this study could be used for selection of maize germplasm with high seed storability.

plant biology↗